resolving the verifybamid results with freemix < 2% and chipmix > 10%

[kiran girdhar]

I ran verifyBAMID on each 200 chip-seq BAM files with input merged vcf file of Genotypes of 600 samples ( 200 of them are chip-seq individuals). I had AF,AC,AN in the info column which was calculated using these genotypes.

~80% of 200 samples are Caucasians.

I am struggling to decide for the samples which have freemix < 2% but chipmix >2%

for sample 1365, I had histone modified chip-seq sample 1365_A with two input samples 1365_Pos_IN_2, 1365_Pos_IN_1. 

1. histone modified chip-seq BAM 1365_A has .00669  whereas chipmix is 0.3765 where with input samples it does resonable job with correct matching of genotype column and reasonable chipmix and freemix values.

Why chipmix is so high for histone sample

#SEQ_ID RG CHIP_ID #SNPS #READS AVG_DP FREEMIX FREELK1 FREELK0 FREE_RH FREE_RA CHIPMIX CHIPLK1 CHIPLK0 CHIP_RH CHIP_RA DPREF RDPHET RDPALT

1365_A ALL 1365_A 683278 436495 0.64 0.00669 160631.69 160867.97 NA NA 0.37645 145839.46 194576.92 NA NA 0.657 0.9478 0.9369

1365_Pos_IN_2 ALL 1365_A 683278 1770725 2.59 0.00320 676736.66 676773.49 NA NA 0.00298 425582.72 426459.54 NA NA 2.598 0.9965 0.9923

1365_Pos_IN_1 ALL 1365_A 683278 1127574 1.65 0.00000 446421.43 446421.43 NA NA 0.00193 254020.46 254353.15 NA NA 1.653 0.9961 0.9982

2. What does it mean when your target sample matched with the intended sample genotype from vcf file with < 2% free mix value but with chipmix value > 2%

for sample 1368_A and 1213_A

1368_A ALL 1368_A 683278 450800 0.66 0.01019 167252.18 167673.56 NA NA 0.56004 158591.79 248605.77 NA NA 0.677 0.9507 0.9477

1213_A ALL 1213_A 683278 945562 1.38 0.01361 345881.30 347208.39 NA NA 0.97533 346470.85 647079.17 NA NA 1.396 0.9783 0.9891

3. This is clearly contaminated sample with 1287_A. then why its chipmix is not 0.9?

1215_Pos_IN ALL 1287_A 683278 3336277 4.88 0.15412 1360018.71 1407231.58 NA NA 0.18590 1109098.83 1431053.16 NA NA 4.900 0.9941 0.9917

[Hyun Min Kang]

Dear Kiran,

For second problem, if FREEMIX is 3% and CHIPMIX is 97%, for example, it means that the intended sample has only 3% or reads while contaminating sample is 97%, suggesting sample swap and contamination.

For ChiP-seq data, it matters how to select the markers to accurately estimate contamination. Could you describe a bit more how you selected the VCF markers and genotypes, and how the reads are aligned, and what the read length is, so on?

Hyun.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.
To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[kiran girdhar]

We have a 683278 markers in input vcf file after maf filter of .05. I haven't subsetted to any specific markers. Our data is from neurons collected from brain. 

Samples were genotyped on the Illumina Infinium HumanOmniExpressExome array (958,178 single-nucleotide polymorphisms, or SNPs) and imputed using standard techniques with the 1,000 Genomes Project as reference data11. 

read are aligned using BWA algorithm. every file has ~2% duplication rate. The read length is 75 base pair.

these are H3K4me3, H3K27ac files.

the cohort is mainly ~80% caucasians

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

[Hyun Min Kang]

Did you set max depth parameter or used the default option? Can you check what happens when you change the parameter? Note that you need to turn --precise option when increasing max depth to something over 30

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

[kiran girdhar]

Hi Hyun

the maxdepth parameters doesn't show a lot of improvement in chipmix value. I don't these ChipMix Values. Freemix value is low which says bam has low contamination but chipmix is capturing what here? it does show the match with the right genotype individual 1365_A.

--maxDepth [30], --precise [ON]

1365_A ALL 1365_A 683278 508979 0.74 0.00498 177267.09 177524.46 NA NA 0.37083 162674.98 211159.66 NA NA 0.770 0.9399 0.9292

--maxDepth [40], --precise [ON]

1365_A ALL 1365_A 683278 571239 0.84 0.00482 191425.74 191742.11 NA NA 0.36799 176940.54 225334.76 NA NA 0.866 0.9334 0.9245

--maxDepth [50], --precise [ON]

1365_A ALL 1365_A 683278 626262 0.92 0.00412 203817.00 204127.67 NA NA 0.36676 189410.79 237716.80 NA NA 0.952 0.9270 0.9208

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

[kiran girdhar]

 

Here is the result for one more sample. and similar kinda trend.

 --maxDepth [20]

1377_A ALL 1377_A 683278 397015 0.58 0.00394 122669.06 122848.53 NA NA 0.28071 112568.21 127407.10 NA NA 0.599 0.9241 0.9657

  --maxDepth [30] --precise[ON]

1377_A ALL 1377_A 683278 474739 0.69 0.00323 138916.59 139138.96 NA NA 0.28463 128924.56 143613.05 NA NA 0.721 0.9109 0.9574

  --maxDepth [40]  --precise[ON]

1377_A ALL 1377_A 683278 540653 0.79 0.00323 152502.07 152783.44 NA NA 0.27613 142585.67 157244.56 NA NA 0.825 0.9011 0.9495

  --maxDepth [50]  --precise[ON]

1377_A ALL 1377_A 683278 597967 0.88 0.00312 164252.11 164581.86 NA NA 0.26976 154416.05 169025.98 NA NA 0.915 0.8929 0.9437

  --maxDepth [60]  --precise[ON]

1377_A ALL 1377_A 683278 648477 0.95 0.00310 174369.85 174722.58 NA NA 0.26769 164592.29 179164.26 NA NA 0.996 0.8846 0.9389

  --maxDepth [70]  --precise[ON]

1377_A ALL 1377_A 683278 693128 1.01 0.00305 183195.98 183579.12 NA NA 0.26719 173456.40 188013.60 NA NA 1.068 0.8777 0.9353

KG

[Hyun Min Kang]

There might be three potential reasons to explain this:

First, the sequence may have very high duplication rates, and lower the heterozygosity systematically, making freemix underestimated. But the difference seem too large to explain this.

Second, there are confounding with contamination and depth. For example, lower depth marker is enriched for contaminated sample while highet depth marker is not. This happens often to ChiPseq samples when only background of foreground sample is contaminated.

Third possibility is that GWAS data is incorrect for some reason. 

Does any of these scenario make sense to you?

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

[kiran girdhar]

1: duplication rate is < 5% 

2. "lower depth marker is enriched for contaminated samples while highest depth marker is not. This happens often to ChiPseq samples when only background of foreground sample is contaminated."

 if this is true I would have to call these samples as contaminate. these high values of chipmix are not due to the verifybamid algorithm limitations?

do you have any suggestion to check the contamination of samples with chipmix values with some other technique?

3. GWAS data is incorrect for some reason. I don't get this

also, a Not on bam files depth issue

I took my vcf file and converted to bed file as chr, start_position, start_positon+1. Then I ran

bedtools coverage -a 1147_A.bam -b my_vcf_bed > output.bed

I took the thirteen column which is coverage depth | sort | uniqued 

Thats how the counts (col 1) and coverage depth(col2) look. 

3901126       .

109725412  0

  2023142      1

  104472       2

   4761          3

    752            4

     35             5

      8               6

      1                7

our genotype vcf file of samples doesn't have overlap with chip-seq bam files in order to get coverage depth. You are right. What should be the expected number there? ~30

May be thats why chipmix values don't make any sense. I have WGS file of some individuals but not for all samples. what is your suggestion for sample swap/ contamination check for those samples?

__________ 

KG

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

[Hyun Min Kang]

I usually trust chipmix than freemix of theu disagree. Freemix can use only the sites with depth 2 or greater while chipmix can use depth 1 sites as well, so if nothing was wrong with the results, natural interpretation would be higher depth sites are not contaminated (i.e. no excess heterozygosity), but lower depth sites show discordances with array data at a certain degree. Do you agree with this interpretation or would their somethinf else be wrong?

If you have background and foreground data as separate library, you could verify this by looking at per library output.

Hyun.

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

[kiran girdhar]

I see your point.

I do have background of some samples like 30 out of 523. eg. see the discrepancy in chipmix values below. they were preapred on the same day and were in the same batch.

I don't get exactly what do you mean by this "If you have background and foreground data as separate library, you could verify this by looking at per library output"

if I set chipmix >2 %, i get 139 out of 307 samples contaminated. Not sure how much to believe. also its kinda weird that all low depth sites are contaminated in these contaminated samples. 

In the end, my goal is to remove bad samples and call peaks on these samples. I usually combine all background samples into one and use that as a one background in macs2.

if the contaminated sites are of low depth then they shouldn't be captured by peaks. right?

foreground   1365_A ALL 1365_A 683278 436495 0.64 0.00669 160631.69 160867.97 NA NA 0.37645 145839.46 194576.92 NA NA 0.657 0.9478 0.9369

background- Library-2 1365_Pos_IN_2 ALL 1365_A 683278 1770725 2.59 0.00320 676736.66 676773.49 NA NA 0.00298 425582.72 426459.54 NA NA 2.598 0.9965 0.9923

background- Library-1 1365_Pos_IN_1 ALL 1365_A 683278 1127574 1.65 0.00000 446421.43 446421.43 NA NA 0.00193 254020.46 254353.15 NA NA 1.653 0.9961 0.9982

KG

Hyun.

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

[Hyun Min Kang]

Chip-seq data have background and peak data sequenced in different library but combined in the same BAM file sometimes. If that waa the case, you can look at per library group results to see if there are within-library contamination. Otherwise, never mind

Hyun.

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

[kiran girdhar]

No we did not combine the data. We sequenced the library separately. 

Also, what’s your suggestion on resolving this contaminated low depth reads in so many samples.

I am scared now if that is true.

Hyun.

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
KG

[Hyun Min Kang]

I would make sure first that the sample is highly contaminated. With that much contamination, you should be able to compare whether you have many homref genotypes containing many alt alleles or many homalt genotypes containing ref alleles.

I also try to estimate FREEMIX again with different sets of markers. I wonder what happens if you use VCF provided by default with verifyBamID. I still am surprised how FREEMIX could be so low, if they are just single library.

Also, make sure that genome build matches with BAM and VCF.

If there is a strong evidence that the sample is indeed contami nated, there are very few things you can do to rescue the sample with such high contamination

Hyun.

Hyun.

KG

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
KG

--

You received this message because you are subscribed to the Google Groups "verifyBamID" group.

To unsubscribe from this group and stop receiving emails from it, send an email to verifybamid...@googlegroups.com.